In this paper, we propose a new approach to solve the magnetostatic inverse problem. The goal of the work is, from measurements of the magnetic field in the air, to rebuild a model for the magnetization of a ferromagnetic shell structure. It's then possible to calculate the field where sensors cannot be placed. This problem is usually ill posed or rank-deficient, it's then necessary to use mathematical regularizations. These techniques are based upon the injection of knowledge about the mathematical behavior of the solution. We preferred to add physical information. This solution allows us to get a faithful solution and to reduce significantly the number of sensors. Moreover, our method has been tested on a mock-up with real measurements and led to very promising results.

This paper proposes a review of the magnetostatic moments method (MoM) applied to model electromagnetic devices. This method is now well-known for its "light weight" and its simplicity of implementation. Its main advantages are the nonrequirement of an air region mesh and a coarse mesh of the ferromagnetic material. It leads to very fast resolution and very accurate field, force, and moment computations. The paper proposes a state of the art of this approach and shows some efficient realizations.Index Terms-Magnetization identification, magnetostatic moment method, point-matching approach, simplified moment method.

No abstract

Abstract-This paper presents an original approach for determining the unknown magnetization of a ferromagnetic shell. Magnetic measurements using sensors close from the device under test are used to rebuild distributions located on the shell. These distributions are representative of the magnetization and tangential moments or charges can be used. This identification problem is a particular case of an inverse problem and is generally ill posed. Instead of using classical mathematical tools to solve such a problem, we preferred to change it in a better posed one by adding our physical knowledge of the problem. All our results have been validated on a mockup with real measurements.

This paper presents a novel and useful 3D nonlinear magnetostatic integral formulation for volume integral method. Like every other integral formulation, its main advantage is that it does not require air region mesh, only ferromagnetic regions being discretized. The formulation is based on magnetic flux density interpolation on facet elements. Special care is taken in order to accurately compute the singularity of Green's kernel. The application of an equivalent circuit approach allows preserving the solenoidality of magnetic induction. It is shown that the formulation is very accurate even if it is associated with coarse meshes. Thus, computation time can be very competitive. Computed results for the TEAM Workshop problem 13 and for a multiply-connected regions case-test are reported.

This paper presents an adapted partial element equivalent circuit (PEEC)-based methodology applied to the modeling of interconnections of power electronics devices. Although this method is already well known, the originality of this work is its use to model a device presenting an industrial complexity. To make possible this modeling, two adapted integral methods, based on two different meshings, are presented. They are dedicated respectively to the computation of parasitic inductances and capacitances and lead to an equivalent circuit of the system. From a time-domain simulation of this circuit, current and voltage sources can be extracted and used to compute the radiated near magnetic field. This approach has been applied to model a real industrial static converter via system couplings, a variable speed drive. Good agreements have been obtained between simulated and measured results on conducted and emitted electromagnetic analysis.Index Terms-Electromagnetic compatibility, fast multipole method, parasitic capacitances, parasitic elements, partial element equivalent circuit (PEEC), power electronics, power interconnections.

A volume integral formulation to compute eddy currents in non-magnetic conductive media is presented. The current distribution is approximated with facet finite elements. The formulation is general and leads to an equivalent lumped elements circuit. In order to ensure the solenoidality of the current distribution, an algorithm detecting the independent loops is then used for the resolution. The formulation is tested on TEAM workshop Problem 7. Even with coarse meshes, its accuracy is demonstrated.

A new integral formulation is presented, enabling the computation of resistive, inductive, and capacitive effects considering both conductors and dielectrics in the frequency domain. The considered application allows us to neglect any propagation effects and magnetic materials. In this paper, we will show how to improve the unstructured-partial element equivalent circuit approach to consider dielectric materials, keeping the same benefits. Results obtained with this formulation are compared to results from an industrial finite-element method software and measurements.